cdec cfg 'cdec.ini' Loading the LM will be faster if you build a binary file. Reading ../standard//nc-wmt11.en.srilm.gz ----5---10---15---20---25---30---35---40---45---50---55---60---65---70---75---80---85---90---95--100 **************************************************************************************************** Seeding random number sequence to 4126799437 dtrain Parameters: k 100 N 4 T 1 scorer 'stupid_bleu' sample from 'kbest' filter 'uniq' learning rate 0.0001 gamma 0 loss margin 1 faster perceptron 0 pairs 'XYX' hi lo 0.1 pair threshold 0 select weights 'last' l1 reg 0 'none' max pairs 4294967295 cdec cfg 'cdec.ini' input 'work/shard.1.0.in' refs 'work/shard.1.0.refs' output 'work/weights.1.0' (a dot represents 10 inputs) Iteration #1 of 1. 5 WEIGHTS Glue = -0.3815 WordPenalty = +0.20064 LanguageModel = +0.95304 LanguageModel_OOV = -0.264 PhraseModel_0 = -0.22362 PhraseModel_1 = +0.12254 PhraseModel_2 = +0.26328 PhraseModel_3 = +0.38018 PhraseModel_4 = -0.48654 PhraseModel_5 = +0 PhraseModel_6 = -0.3645 PassThrough = -0.2216 --- 1best avg score: 0.10863 (+0.10863) 1best avg model score: -4.9841 (-4.9841) avg # pairs: 1345.4 avg # rank err: 822.4 avg # margin viol: 501 non0 feature count: 11 avg list sz: 100 avg f count: 11.814 (time 0.43 min, 5.2 s/S) Writing weights file to 'work/weights.1.0' ... done --- Best iteration: 1 [SCORE 'stupid_bleu'=0.10863]. This took 0.43333 min.